Flood hazard model calibration using multiresolution model output
share
Riverine floods pose a considerable risk to many communities. Improving the projections of flood hazard has the potential to inform the design and implementation of flood risk management strategies. Current flood hazard projections are uncertain. One uncertainty that is often overlooked is uncertainty about model parameters. Calibration methods use observations to quantify model parameter uncertainty. With limited computational resources, researchers typically calibrate models using either relatively few expensive model runs at a high spatial resolution or many cheaper runs at a lower spatial resolution. This leads to an open question: Is it possible to effectively combine information from the high and low resolution model runs? We propose a Gaussian process-based Bayesian emulation-calibration approach that assimilates model outputs and observations at multiple resolutions. We demonstrate our approach using the LISFLOOD-FP flood hazard model as a case study for a riverine community in Pennsylvania in the Eastern United States. Compared to considered existing single-resolution approaches, our method yields more accurate flood predictions. Our method is rather general and can be applied to calibrate other high dimensional computer models to help improve future projections.
READ FULL TEXT
